Lihua Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Lihua Li has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Jian Zhou
School of Mechanical Engineering, Hefei University of Technology, Hefei 230009, China
6 shared publications - Xingzheng Wang
College of Mechatronics and Control Engineering, Shenzhen University, Shenzhen, China
3 shared publications Kun Liu
School of Mechanical Engineering Hefei University of Technology Hefei Anhui China
2 shared publications- Man Cheung Ng
The State Key Laboratory of Ultraprecision Machining Technology Department of Industrial and Systems Engineering The Hong Kong Polytechnic University Kowloon Hong Kong
2 shared publications - Wing Bun Lee
The Hong Kong Polytechnic University
2 shared publications Wing-Bun Lee
State Key Laboratory of Ultraprecision Machining Technology, Department of Industrial and Systems Engineering, Hong Kong Polytechnic University, Kowloon, Hong Kong
2 shared publications
Latest Publications
- Open Access
Tribological Properties and Wear Mechanisms of Carbide-Bonded Graphene Coating on Silicon Substrate
Carbide-bonded graphene (CBG) coating, with its unique 3D cross-linked network structure, shows significant potential for protecting silicon substrates. However, a comprehensive understanding of its macroscale tribological properties remains lacking. This study investigated the macroscale friction and wear behaviors of CBG-coated silicon wafers using reciprocating sliding tests against steel balls under various loads and sliding cycles. The CBG coating exhibited excellent friction-reduction and anti-wear performance, reducing the steady friction coefficient from 0.80 to 0.17 and wear rate by an order of magnitude compared to those of bare silicon. Higher loads slightly decreased both friction coefficients and wear rates, primarily due to the formation of denser tribofilms and transfer layers. Re-running experiments revealed three distinct wear stages—adhesive, abrasive, and accelerated substrate wear—driven by the evolution of tribofilms, transfer layers, and unabraded flat areas. Furthermore, comparative experiments confirmed that these “unabraded flat areas” on the wear track play a critical role in sustaining low friction and prolonging coating life. The findings identify CBG as a robust solid lubricant for high-contact-pressure applications and emphasize the influence of tribo-layer dynamics and wear debris behavior on coating performance.
Open AccessTribological Properties and Wear Mechanisms of Carbide-Bonded Graphene Coating on Silicon Substrate
Carbide-bonded graphene (CBG) coating, with its unique 3D cross-linked network structure, shows significant potential for protecting silicon substrates. However, a comprehensive understanding of its macroscale tribological properties remains lacking. This study investigated the macroscale friction and wear behaviors of CBG-coated silicon wafers using reciprocating sliding tests against steel balls under various loads and sliding cycles. The CBG coating exhibited excellent friction-reduction and anti-wear performance, reducing the steady friction coefficient from 0.80 to 0.17 and wear rate by an order of magnitude compared to those of bare silicon. Higher loads slightly decreased both friction coefficients and wear rates, primarily due to the formation of denser tribofilms and transfer layers. Re-running experiments revealed three distinct wear stages—adhesive, abrasive, and accelerated substrate wear—driven by the evolution of tribofilms, transfer layers, and unabraded flat areas. Furthermore, comparative experiments confirmed that these “unabraded flat areas” on the wear track play a critical role in sustaining low friction and prolonging coating life. The findings identify CBG as a robust solid lubricant for high-contact-pressure applications and emphasize the influence of tribo-layer dynamics and wear debris behavior on coating performance.
- Open Access
- 30 May 2025
- Published by MDPI AG in Micromachines
- Vol. 16 (6) , 660
The Principle and Application of Achromatic Metalens
Metalenses, as ultrathin planar optical devices based on metasurfaces, have attracted significant research interest in recent years due to their compact structure and versatile light manipulation, showing great potential to replace traditional lenses in specific applications. This review focuses on the fundamental principles of geometric and propagation phases of metalenses, introducing their applications in achromatic aberration correction, and emphasizing their advantages and limitations. We further discuss the application of multilayer metalenses in zoom optical systems and summarize methods such as topology optimization and inverse design to enhance the efficiency of metalenses. Special attention is given to comparing broadband and discrete achromatic correction, highlighting their respective working principles, design challenges, and practical implications. Additionally, recent advances in using deep learning for image-side aberration correction are briefly discussed. Finally, we highlight various practical applications of metalenses and discuss future research directions.
- 30 May 2025
- Published by MDPI AG in Micromachines
- Vol. 16 (6) , 660
Open AccessThe Principle and Application of Achromatic Metalens
Metalenses, as ultrathin planar optical devices based on metasurfaces, have attracted significant research interest in recent years due to their compact structure and versatile light manipulation, showing great potential to replace traditional lenses in specific applications. This review focuses on the fundamental principles of geometric and propagation phases of metalenses, introducing their applications in achromatic aberration correction, and emphasizing their advantages and limitations. We further discuss the application of multilayer metalenses in zoom optical systems and summarize methods such as topology optimization and inverse design to enhance the efficiency of metalenses. Special attention is given to comparing broadband and discrete achromatic correction, highlighting their respective working principles, design challenges, and practical implications. Additionally, recent advances in using deep learning for image-side aberration correction are briefly discussed. Finally, we highlight various practical applications of metalenses and discuss future research directions.
- 20 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (6) , 10272-10286
DADAN: Dynamic-Augmented and Density-Aware Network for Accurate 3-D Object Detection With 4-D Radar
The cost-effective automotive 4D radar is emerging as a feasible solution for 3D object detection in autonomous driving scenarios. However, point clouds generated by 4D radar are typically sparse. Current methods encounter challenges in resolving the foreground-background confusion in sparse point clouds due to (1) the underutilization of the unique dynamic information (velocity and intensity) of 4D radar points, and (2) the density variations of point clouds where they are near-dense-far-sparse, exacerbating the confusion. Therefore, we propose a Dynamic-Augmented and Density-Aware Network (DADAN) to address these challenges by: (1) assigning high weights to foreground points based on dynamic information and enhancing their feature encoding to reduce confusion effectively, and (2) utilizing attention mechanisms along with density information to adaptively address exacerbated confusion caused by density variations. DADAN comprises a dynamic-augmented encoding network, a density-aware backbone, and a detection head. The encoding network incorporates a Dynamic-Augmented Encoding Module (DAEM) that assigns high weights to foreground points based on velocity and intensity data to improve their feature encoding. Notably, foreground points typically exhibit distinct dynamic information compared to background points, distinguishing them from the background. The backbone network incorporates an attention mechanism along with density information to generate density-aware features. The detection head is responsible for 3D bounding boxes and object categories. The experimental results on various public datasets demonstrate the superior performance of the proposed method compared to existing approaches. These results illustrate that DADAN effectively addressing the issue of confusion, accurately detecting objects in complex scenes.
- 20 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (6) , 10272-10286
DADAN: Dynamic-Augmented and Density-Aware Network for Accurate 3-D Object Detection With 4-D Radar
The cost-effective automotive 4D radar is emerging as a feasible solution for 3D object detection in autonomous driving scenarios. However, point clouds generated by 4D radar are typically sparse. Current methods encounter challenges in resolving the foreground-background confusion in sparse point clouds due to (1) the underutilization of the unique dynamic information (velocity and intensity) of 4D radar points, and (2) the density variations of point clouds where they are near-dense-far-sparse, exacerbating the confusion. Therefore, we propose a Dynamic-Augmented and Density-Aware Network (DADAN) to address these challenges by: (1) assigning high weights to foreground points based on dynamic information and enhancing their feature encoding to reduce confusion effectively, and (2) utilizing attention mechanisms along with density information to adaptively address exacerbated confusion caused by density variations. DADAN comprises a dynamic-augmented encoding network, a density-aware backbone, and a detection head. The encoding network incorporates a Dynamic-Augmented Encoding Module (DAEM) that assigns high weights to foreground points based on velocity and intensity data to improve their feature encoding. Notably, foreground points typically exhibit distinct dynamic information compared to background points, distinguishing them from the background. The backbone network incorporates an attention mechanism along with density information to generate density-aware features. The detection head is responsible for 3D bounding boxes and object categories. The experimental results on various public datasets demonstrate the superior performance of the proposed method compared to existing approaches. These results illustrate that DADAN effectively addressing the issue of confusion, accurately detecting objects in complex scenes.
- Open Access
- 22 May 2024
- Published by MDPI AG in Micromachines
- Vol. 15 (6) , 673
Modeling and Experiments on Temperature and Electrical Conductivity Characteristics in High-Temperature Heating of Carbide-Bonded Graphene Coating on Silicon
A novel non-isothermal glass hot embossing system utilizes a silicon mold core coated with a three-dimensional carbide-bonded graphene (CBG) coating, which acts as a thin-film resistance heater. The temperature of the system significantly influences the electrical conductivity properties of silicon with a CBG coating. Through simulations and experiments, it has been established that the electrical conductivity of silicon with a CBG coating gradually increases at lower temperatures and rapidly rises as the temperature further increases. The CBG coating predominantly affects electrical conductivity until 400 °C, after which silicon becomes the dominant factor. Furthermore, the dimensions of CBG-coated silicon and the reduction of CBG coating also affect the rate and outcome of conductivity changes. These findings provide valuable insights for detecting CBG-coated silicon during the embossing process, improving efficiency, and predicting the mold core’s service life, thus enhancing the accuracy of optical lens production.
- 22 May 2024
- Published by MDPI AG in Micromachines
- Vol. 15 (6) , 673
Open AccessModeling and Experiments on Temperature and Electrical Conductivity Characteristics in High-Temperature Heating of Carbide-Bonded Graphene Coating on Silicon
A novel non-isothermal glass hot embossing system utilizes a silicon mold core coated with a three-dimensional carbide-bonded graphene (CBG) coating, which acts as a thin-film resistance heater. The temperature of the system significantly influences the electrical conductivity properties of silicon with a CBG coating. Through simulations and experiments, it has been established that the electrical conductivity of silicon with a CBG coating gradually increases at lower temperatures and rapidly rises as the temperature further increases. The CBG coating predominantly affects electrical conductivity until 400 °C, after which silicon becomes the dominant factor. Furthermore, the dimensions of CBG-coated silicon and the reduction of CBG coating also affect the rate and outcome of conductivity changes. These findings provide valuable insights for detecting CBG-coated silicon during the embossing process, improving efficiency, and predicting the mold core’s service life, thus enhancing the accuracy of optical lens production.
- 6 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 9 (25730436) , 1139-1151
Boosting Light Field Image Super Resolution Learnt From Single-Image Prior
In recent years, many deep learning networks are proposed for light field super resolution (LFSR). LFSR problem is essentially ill-posed since unknown detail information need to be predicted. Hence LFSR networks require plentiful content information (e.g., shape, color, texture) learned from sufficiently diverse scenarios. However, due to the high collection cost, existing light field datasets are in small size and have few scenarios, which could not meet the requirement and limit the performance of LFSR networks. To solve this problem, we proposed a novel framework to significantly boost their performance. Our main idea is to introduce valuable and plentiful content information from single images into LFSR networks as prior. Specifically, first, a view synthesis method is applied to add unreal disparity into single images, increasing the dimensionality of single images, hence to solve the problem of inconsistent data modalities. Then, we design the Scenarios-Content Introduction Module (SCIM) to effectively extract content feature from synthesized data. Finally, due to the added unreal disparity in the first stage, the features have severely pseudo information. Hence the Feature Attention Module (FAM) is proposed to discriminately select valuable information and combine this information into the LFSR network. Extensive experiments on six datasets validate the effectiveness of the proposed method, leading to a maximum gain of 0.439 dB. Our method can even boost SOTA networks to achieve higher performance
- 6 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 9 (25730436) , 1139-1151
Boosting Light Field Image Super Resolution Learnt From Single-Image Prior
In recent years, many deep learning networks are proposed for light field super resolution (LFSR). LFSR problem is essentially ill-posed since unknown detail information need to be predicted. Hence LFSR networks require plentiful content information (e.g., shape, color, texture) learned from sufficiently diverse scenarios. However, due to the high collection cost, existing light field datasets are in small size and have few scenarios, which could not meet the requirement and limit the performance of LFSR networks. To solve this problem, we proposed a novel framework to significantly boost their performance. Our main idea is to introduce valuable and plentiful content information from single images into LFSR networks as prior. Specifically, first, a view synthesis method is applied to add unreal disparity into single images, increasing the dimensionality of single images, hence to solve the problem of inconsistent data modalities. Then, we design the Scenarios-Content Introduction Module (SCIM) to effectively extract content feature from synthesized data. Finally, due to the added unreal disparity in the first stage, the features have severely pseudo information. Hence the Feature Attention Module (FAM) is proposed to discriminately select valuable information and combine this information into the LFSR network. Extensive experiments on six datasets validate the effectiveness of the proposed method, leading to a maximum gain of 0.439 dB. Our method can even boost SOTA networks to achieve higher performance
- Open Access
- 4 December 2021
- Published by MDPI AG in Micromachines
- Vol. 12 (12) , 1510
Modeling and Simulation of the Surface Generation Mechanism of a Novel Low-Pressure Lapping Technology
Aluminum alloy (Al6061) is a common material used in the ultraprecision area. It can be machined with a good surface finish by single-point diamond turning (SPDT). Due to the material being relatively soft, it is difficult to apply post-processing techniques such as ultraprecision lapping and ultraprecision polishing, as they may scratch the diamond-turned surface. As a result, a novel low-pressure lapping method was developed by our team to reduce the surface roughness. In this study, a finite element model was developed to simulate the mechanism of this novel lapping technology. The simulation results were compared with the experimental results so as to gain a better understanding of the lapping mechanism.
- 4 December 2021
- Published by MDPI AG in Micromachines
- Vol. 12 (12) , 1510
Open AccessModeling and Simulation of the Surface Generation Mechanism of a Novel Low-Pressure Lapping Technology
Aluminum alloy (Al6061) is a common material used in the ultraprecision area. It can be machined with a good surface finish by single-point diamond turning (SPDT). Due to the material being relatively soft, it is difficult to apply post-processing techniques such as ultraprecision lapping and ultraprecision polishing, as they may scratch the diamond-turned surface. As a result, a novel low-pressure lapping method was developed by our team to reduce the surface roughness. In this study, a finite element model was developed to simulate the mechanism of this novel lapping technology. The simulation results were compared with the experimental results so as to gain a better understanding of the lapping mechanism.
- Open Access
- 13 May 2021
- Published by MDPI AG in Micromachines
- Vol. 12 (5) , 557
Fast and Accurate Light Field View Synthesis by Optimizing Input View Selection
There is a trade-off between spatial resolution and angular resolution limits in light field applications; various targeted algorithms have been proposed to enhance angular resolution while ensuring high spatial resolution simultaneously, which is also called view synthesis. Among them, depth estimation-based methods can use only four corner views to reconstruct a novel view at an arbitrary location. However, depth estimation is a time-consuming process, and the quality of the reconstructed novel view is not only related to the number of the input views, but also the location of the input views. In this paper, we explore the relationship between different input view selections with the angular super-resolution reconstruction results. Different numbers and positions of input views are selected to compare the speed of super-resolution reconstruction and the quality of novel views. Experimental results show that the speed of the algorithm decreases with the increase of the input views for each novel view, and the quality of the novel view decreases with the increase of the distance from the input views. After comparison using two input views in the same line to reconstruct the novel views between them, fast and accurate light field view synthesis is achieved.
- 13 May 2021
- Published by MDPI AG in Micromachines
- Vol. 12 (5) , 557
Open AccessFast and Accurate Light Field View Synthesis by Optimizing Input View Selection
There is a trade-off between spatial resolution and angular resolution limits in light field applications; various targeted algorithms have been proposed to enhance angular resolution while ensuring high spatial resolution simultaneously, which is also called view synthesis. Among them, depth estimation-based methods can use only four corner views to reconstruct a novel view at an arbitrary location. However, depth estimation is a time-consuming process, and the quality of the reconstructed novel view is not only related to the number of the input views, but also the location of the input views. In this paper, we explore the relationship between different input view selections with the angular super-resolution reconstruction results. Different numbers and positions of input views are selected to compare the speed of super-resolution reconstruction and the quality of novel views. Experimental results show that the speed of the algorithm decreases with the increase of the input views for each novel view, and the quality of the novel view decreases with the increase of the distance from the input views. After comparison using two input views in the same line to reconstruct the novel views between them, fast and accurate light field view synthesis is achieved.
- Open Access
- 1 February 2021
- Published by MDPI AG in Nanomaterials
- Vol. 11 (2) , 363
Evaluation of Warpage and Residual Stress of Precision Glass Micro-Optics Heated by Carbide-Bonded Graphene Coating in Hot Embossing Process
A newly developed hot embossing technique which uses the localized rapid heating of a thin carbide-bonded graphene (CBG) coating, greatly reduces the energy consumption and promotes the fabrication efficiency. However, because of the non-isothermal heat transfer process, significant geometric deviation and residual stress could be introduced. In this paper, we successfully facilitate the CBG-heating-based hot embossing into the fabrication of microlens array on inorganic glass N-BK7 substrate, where the forming temperature is as high as 800 °C. The embossed microlens array has high replication fidelity, but an obvious geometric warpage along the glass substrate also arises. Thermo-mechanical coupled finite element modelling of the embossing process is conducted and verified by the experimental results. Based on trial and error simulations, an appropriate compensation curvature is determined and adopted to modify the geometrical design of the silicon wafer mold. The warpage of the re-embossed microlens array is significantly decreased using the compensated mold, which demonstrates the feasibility of the simulation-oriented compensation scheme. Our work would contribute to improving the quality of optics embossed by this innovative CBG-heating-based hot embossing technique.
- 1 February 2021
- Published by MDPI AG in Nanomaterials
- Vol. 11 (2) , 363
Open AccessEvaluation of Warpage and Residual Stress of Precision Glass Micro-Optics Heated by Carbide-Bonded Graphene Coating in Hot Embossing Process
A newly developed hot embossing technique which uses the localized rapid heating of a thin carbide-bonded graphene (CBG) coating, greatly reduces the energy consumption and promotes the fabrication efficiency. However, because of the non-isothermal heat transfer process, significant geometric deviation and residual stress could be introduced. In this paper, we successfully facilitate the CBG-heating-based hot embossing into the fabrication of microlens array on inorganic glass N-BK7 substrate, where the forming temperature is as high as 800 °C. The embossed microlens array has high replication fidelity, but an obvious geometric warpage along the glass substrate also arises. Thermo-mechanical coupled finite element modelling of the embossing process is conducted and verified by the experimental results. Based on trial and error simulations, an appropriate compensation curvature is determined and adopted to modify the geometrical design of the silicon wafer mold. The warpage of the re-embossed microlens array is significantly decreased using the compensated mold, which demonstrates the feasibility of the simulation-oriented compensation scheme. Our work would contribute to improving the quality of optics embossed by this innovative CBG-heating-based hot embossing technique.
- Open Access
Evaluation of a Novel Nanodroplet Cutting Fluid for Diamond Turning of Optical Polymers
In this study, a novel nanodroplet cutting fluid (NDCF), consisting of emulsified water and oil nanodroplets, was developed to improve the surface quality of the single-point diamond-turned optical polymers. This developed NDCF was able to penetrate the chip–tool interface, contributing to both cooling and lubricating effects. The performance of NDCF was evaluated from perspectives of the surface irregularity, roughness, and cutting force of the machined groove in a series of taper cutting experiments. Meanwhile, a high-quality optical surface was obtained and the micro-level form error was reduced in the diamond turning of a Polymethylmethacrylate (PMMA) microlens array.
Open AccessEvaluation of a Novel Nanodroplet Cutting Fluid for Diamond Turning of Optical Polymers
In this study, a novel nanodroplet cutting fluid (NDCF), consisting of emulsified water and oil nanodroplets, was developed to improve the surface quality of the single-point diamond-turned optical polymers. This developed NDCF was able to penetrate the chip–tool interface, contributing to both cooling and lubricating effects. The performance of NDCF was evaluated from perspectives of the surface irregularity, roughness, and cutting force of the machined groove in a series of taper cutting experiments. Meanwhile, a high-quality optical surface was obtained and the micro-level form error was reduced in the diamond turning of a Polymethylmethacrylate (PMMA) microlens array.
- 21 September 2020
- Published by Wiley in Journal of the American Ceramic Society
- Vol. 104 (1) , 243-255
Temperature and rate dependent debonding behaviors of precision glass molding interface
The glass‐to‐mold adhesion in precision glass molding could severely degrade the quality of molded optics and shorten the lifespan of the precious molds. Since the consequences of adhesion take effect during the separation between glass and molds, it is important to investigate the debonding behaviors of a typical glass molding interface. To this end, here we perform a probe tack test procedure for borosilicate glass BK7, where debonding is conducted at molding temperature and specific velocity. We fully characterize the debonding behaviors using the peak adhesion stress σ max and the work of debonding W deb . Experiments show that when temperature is decreased from 690°C to 655°C at 10 μm/s, σ max continuously increases, while W deb first increases but then sharply decreases. When the debonding velocity is increased from 10 to 50 μm/s at 680°C, σ max also increases while W deb overall decreases. Therefore, the debonding behaviors are highly temperature and rate dependent. More importantly, depending on the debonding conditions, three debonding types are identified, that is, the cohesive bulk deformation, the cohesive‐interfacial transition and the interfacial fracture. The cohesive type can be converted into the interfacial fracture, by either decreasing temperature or increasing the debonding velocity. Based on the W deb criterion, the three debonding regimes can be clearly distinguished. Finally, analyses on the temperature and velocity experimental results are unified by incorporating the reduced crack velocity a T v c . The dependences of both viscoelasticity and W deb on a T v c qualitatively explain the transition condition for different debonding types. Concerning these findings, the work of debonding not only supplements the characterization of adhesion strength, but also throws insightful light on revealing the debonding mechanisms.
- 21 September 2020
- Published by Wiley in Journal of the American Ceramic Society
- Vol. 104 (1) , 243-255
Temperature and rate dependent debonding behaviors of precision glass molding interface
The glass‐to‐mold adhesion in precision glass molding could severely degrade the quality of molded optics and shorten the lifespan of the precious molds. Since the consequences of adhesion take effect during the separation between glass and molds, it is important to investigate the debonding behaviors of a typical glass molding interface. To this end, here we perform a probe tack test procedure for borosilicate glass BK7, where debonding is conducted at molding temperature and specific velocity. We fully characterize the debonding behaviors using the peak adhesion stress σ max and the work of debonding W deb . Experiments show that when temperature is decreased from 690°C to 655°C at 10 μm/s, σ max continuously increases, while W deb first increases but then sharply decreases. When the debonding velocity is increased from 10 to 50 μm/s at 680°C, σ max also increases while W deb overall decreases. Therefore, the debonding behaviors are highly temperature and rate dependent. More importantly, depending on the debonding conditions, three debonding types are identified, that is, the cohesive bulk deformation, the cohesive‐interfacial transition and the interfacial fracture. The cohesive type can be converted into the interfacial fracture, by either decreasing temperature or increasing the debonding velocity. Based on the W deb criterion, the three debonding regimes can be clearly distinguished. Finally, analyses on the temperature and velocity experimental results are unified by incorporating the reduced crack velocity a T v c . The dependences of both viscoelasticity and W deb on a T v c qualitatively explain the transition condition for different debonding types. Concerning these findings, the work of debonding not only supplements the characterization of adhesion strength, but also throws insightful light on revealing the debonding mechanisms.